Karpenter workflow

Skill iceflower/agent-skills/karpenter-workflow

Karpenter core workflow, NodePool configuration, autoscaling patterns, and troubleshooting. Includes cloud provider configurations (AWS EC2NodeClass, Azure AKSNodeClass, GCP GKENodeClass). Use for Karpenter operations.From its SKILL.md

Install

npx -y skills add iceflower/agent-skills --skill karpenter-workflow

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

10.3 KB, ~2.7k tokens by cl100k_base, as published. Nobody here has run it

Karpenter Workflow Guide

1. Overview

Karpenter is an open-source Kubernetes node autoscaler designed for flexibility, performance, and simplicity. Unlike Cluster Autoscaler, Karpenter provisions nodes based on pod requirements rather than node group configurations.

Key Features

Just-in-time provisioning: Nodes created when pods are pending
Flexible scheduling: Considers CPU, memory, GPU, storage, and topology
Consolidation: Automatically replaces nodes with cheaper alternatives
Multi-cloud support: AWS, Azure, and GKE

Karpenter vs Cluster Autoscaler

Feature	Karpenter	Cluster Autoscaler
Scaling model	Pod-driven	Node group-driven
Instance selection	Dynamic	Pre-configured groups
Spot support	Native	Limited
Consolidation	Active	Passive
Cloud support	AWS, Azure, GKE	All major clouds

2. Version Support (2026-03-14)

Latest Version

Version	Release Date	Kubernetes Compatibility
v1.0.x	Feb 2026	1.31+
v0.37.x	Oct 2025	1.30
v0.34.x	Jul 2025	1.29
v0.31.x	Apr 2025	1.28

Compatibility Matrix

Kubernetes	Karpenter
1.31	v1.0.5+
1.30	v0.37.x
1.29	v0.34.x
1.28	v0.31.x
1.27	v0.28.x
1.25	v0.25.x

3. Core Concepts

NodePool

Defines where and how Karpenter should provision nodes.

apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: default
spec:
  template:
    spec:
      requirements:
        - key: kubernetes.io/arch
          operator: In
          values: ["amd64"]
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["spot", "on-demand"]
      nodeClassRef:
        group: karpenter.k8s.aws
        kind: EC2NodeClass
        name: default
  limits:
    cpu: 1000
    memory: 1000Gi
  disruption:
    consolidationPolicy: WhenEmptyOrUnderutilized
    consolidateAfter: 1m

EC2NodeClass / AKSNodeClass / GKENodeClass

Provider-specific node configuration.

apiVersion: karpenter.k8s.aws/v1
kind: EC2NodeClass
metadata:
  name: default
spec:
  amiSelectorTerms:
    - alias: al2023@latest
  role: KarpenterNodeRole
  subnetSelectorTerms:
    - tags:
        karpenter.sh/discovery: my-cluster
  securityGroupSelectorTerms:
    - tags:
        karpenter.sh/discovery: my-cluster
  tags:
    Environment: production

4. Autoscaling Configuration

Requirements

Define node constraints using Kubernetes label selectors:

Requirement Key	Values	Description
`karpenter.sh/capacity-type`	spot, on-demand	Instance market type
`kubernetes.io/arch`	amd64, arm64	CPU architecture
`kubernetes.io/os`	linux, windows	Operating system
`karpenter.k8s.aws/instance-category`	c, m, r, etc.	Instance family
`karpenter.k8s.aws/instance-generation`	5, 6, 7	Instance generation

spec:
  template:
    spec:
      requirements:
        # Capacity type
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["spot", "on-demand"]

        # Instance types
        - key: node.kubernetes.io/instance-type
          operator: In
          values: ["m6i.xlarge", "m6i.2xlarge", "m7i.xlarge"]

        # Architecture
        - key: kubernetes.io/arch
          operator: In
          values: ["amd64"]

        # Exclude small instances
        - key: karpenter.k8s.aws/instance-size
          operator: NotIn
          values: ["nano", "micro", "small"]

Resource Limits

spec:
  limits:
    cpu: 1000      # Max 1000 vCPUs
    memory: 1000Gi # Max 1000GiB memory

5. Disruption Policies

Consolidation Policies

Policy	Description
`WhenEmpty`	Replace empty nodes only
`WhenEmptyOrUnderutilized`	Replace underutilized nodes (recommended)

spec:
  disruption:
    consolidationPolicy: WhenEmptyOrUnderutilized
    consolidateAfter: 1m
    budgets:
      - nodes: 10%
        schedule: "0 0 * * *"  # Allow 10% disruption at midnight
      - nodes: 0               # No disruption during business hours
        schedule: "0 9-17 * * MON-FRI"

Expiration

spec:
  disruption:
    expireAfter: 720h  # Nodes expire after 30 days

6. Pod Scheduling

Node Affinity

Karpenter considers pod scheduling constraints:

spec:
  affinity:
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
          - matchExpressions:
              - key: topology.kubernetes.io/zone
                operator: In
                values: ["us-east-1a", "us-east-1b"]

Topology Spread

spec:
  topologySpreadConstraints:
    - maxSkew: 1
      topologyKey: topology.kubernetes.io/zone
      whenUnsatisfiable: ScheduleAnyway
      labelSelector:
        matchLabels:
          app: myapp

Priority Classes

apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: high-priority
value: 1000000
globalDefault: false
---
apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: high-priority
spec:
  template:
    spec:
      # Higher priority pods get dedicated nodes

7. Spot Instance Handling

Spot Interruption

Karpenter automatically handles spot interruptions:

2-minute warning: AWS sends spot interruption notice
Cordon and drain: Karpenter cordons node and drains pods
Replacement: New node provisioned before termination

Spot Diversification

spec:
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["spot"]
        - key: karpenter.k8s.aws/instance-family
          operator: In
          values: ["m6i", "m7i", "c6i", "c7i"]  # Multiple families

Spot Fallback

spec:
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["spot", "on-demand"]  # Fall back to on-demand

8. Node Initialization

Init Containers

spec:
  template:
    spec:
      startupTaints:
        - key: node.kubernetes.io/not-ready
          effect: NoSchedule
      initContainers:
        - name: init
          image: busybox
          command: ["/bin/sh", "-c", "echo initializing"]

UserData

spec:
  userData: |
    #!/bin/bash
    echo "Custom node initialization"
    /etc/eks/bootstrap.sh my-cluster

9. Monitoring

Key Metrics

Metric	Description
`karpenter_nodes_created_total`	Total nodes created
`karpenter_nodes_terminated_total`	Total nodes terminated
`karpenter_pods_pending_total`	Pods waiting for nodes
`karpenter_provisioning_duration_seconds`	Time to provision node

Grafana Dashboard

Import Karpenter dashboard: https://grafana.com/grafana/dashboards/1860

Logging

# View Karpenter logs
kubectl logs -n kube-system -l app.kubernetes.io/name=karpenter

# Watch provisioning events
kubectl get events -A --field-selector reason=Provisioning

10. Troubleshooting

Pods Stuck in Pending

# Check unschedulable pods
kubectl get pods -A --field-selector=status.phase=Pending -o wide

# Describe pod for constraints
kubectl describe pod <pod-name>

# Check Karpenter logs
kubectl logs -n kube-system -l app.kubernetes.io/name=karpenter | grep "cannot be scheduled"

Nodes Not Provisioning

# Check NodePool status
kubectl get nodepool -o yaml

# Check EC2NodeClass status
kubectl get ec2nodeclass -o yaml

# Check Karpenter controller
kubectl get pods -n kube-system -l app.kubernetes.io/name=karpenter

Spot Instance Issues

# Check spot capacity
aws ec2 describe-spot-instance-requests

# Review spot interruption history
kubectl get events -A --field-selector reason=SpotInterruption

11. Best Practices

1. Use Multiple NodePools

# Critical workloads - on-demand
---
apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: critical
spec:
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["on-demand"]

---
# Non-critical - spot
apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: spot
spec:
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: ["spot"]

2. Set Resource Limits

Prevent runaway scaling:

spec:
  limits:
    cpu: 1000
    memory: 1000Gi

3. Use Consolidation

Reduce costs by consolidating workloads:

spec:
  disruption:
    consolidationPolicy: WhenEmptyOrUnderutilized
    consolidateAfter: 1m

4. Define Budgets

Control disruption rate:

spec:
  disruption:
    budgets:
      - nodes: 10%

12. Provider-Specific Skills

For provider-specific configurations, see dedicated skills:

AWS EKS: EKS section in karpenter-providers rules
Azure AKS: AKS section in karpenter-providers rules
GCP GKE: GKE section in karpenter-providers rules

13. Migration from Cluster Autoscaler

Steps

Install Karpenter alongside Cluster Autoscaler
Create NodePools matching existing node groups
Gradually reduce Cluster Autoscaler node groups
Remove Cluster Autoscaler

Migration Checklist

14. References

Additional References

For cloud provider configurations (EC2NodeClass, AKSNodeClass, GKENodeClass), see references/providers.md

What ships with it: 2 files

25.4 KB alongside SKILL.md

references/

providers.md24.5 KB

README.md919 B

Gives 0 of the 12 instructions most automation workflows skills give in ~2.7k tokens

Counted across 745 of the 1,008 authors here whose files we hold, read 2026-08-07

Write conventional commit messagesin 36 of 745, across 35 files
Delete branches after mergein 30 of 745, across 21 files
Make atomic commitsin 25 of 745, across 15 files
Write minimal code to pass testsin 22 of 745, across 10 files
Re-snapshot after navigation or DOM changesin 21 of 745, across 13 files
Use try-catch for error handlingin 20 of 745, across 8 files
Run tests before committingin 20 of 745, across 12 files
Write tests before implementationin 20 of 745, across 8 files
Configure branch protection rulesin 19 of 745, across 5 files
Explain the why in commit messagesin 19 of 745, across 9 files
Refactor code while tests remain greenin 19 of 745, across 6 files
Interact with elements using refsin 19 of 745, across 11 files

Said here and by no other author read

Define NodePool requirements using label selectors
Set resource limits to prevent runaway scaling
Use multiple NodePools to isolate workloads
Enable node consolidation for underutilized nodes
Define disruption budgets to control disruption rates
Configure multiple spot instance families for diversification

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.